I. Introduction
The purpose of a fECG is to measure the electrical motion of the fetal cardiac. Fetal ECG is acquired through the mother’s abdomen by keeping ECG probe on the mother’s belly. It is a mixed signal consisting of a mother and fetus ECG signal along with motion artifact, baseline drift, and an electromyogram. But extracting the fetus ECG is a need of today’s era to know the early health condition of the baby. With the development of technology, wearable fetal electronics monitors such as Monica An24 fetal monitor (Nottingham, UK), Nemo Healthcare (Veldhoven, the Netherlands), and Meridian M110 (North Andover, MA, USA) are getting popular and can monitor fetus ECG. Invasive techniques give maximum accuracy. But accurate FECG morphology extraction and fetal cardiac disease detection in a non-invasive way remain an open problem for research. During labor, a fetal heart monitor is used to check fetal health status. However, the current monitoring instrument is based on an ultrasonic technique that also captures uterine contractions and fetal movements [1] [2] . This could lead to a misleading opinion regarding fetal heart rate. This can result in cesarean delivery. As per the national family health survey (NFHS-5) [3] , the national c-section rate in 2020–21, in the private sector is 47.4% and in the government sector is 14.3% whereas the WHO recommends cesarean delivery should not exceed 10–15%. M. Cremer was the first to observe a fetal electrocardiogram in 1906. A galvanometric apparatus is used to check the fetal electrocardiogram, but it had a very low amplitude of the fetal signal. Though the extraction of fetal electrocardiograms is a challenge, researchers are working on finding the best method for extracting fetal ECGs in a non-invasive way. Extraction of fetal electrocardiogram from multi-channel recording is possible through independent component analysis [4] , [5] , [6] . R. Kahankova et. al. [7] demonstrated the Variation of fetal heart rate with gestational age using the RMS and LMS algorithms. The author also mentioned that fetal heart rate can be captured from 8 weeks of pregnancy. The RMS algorithm is most suitable at high gestation age i.e. 30th week on a ward, and adaptive filtering with LMS algorithm is best suited at an earlier stage of pregnancy. Echo state neural network (ESN)-based filtering is used to cancel the MECG component and suggest a unique technique for FECG extraction [8] . The least mean squares (LMS) and recursive least squares (RLS) algorithms are used to minimize and calculate the error function between the expected output and the actual output of the adaptive algorithm [5] . The adaptive neuro-fuzzy interference system (ANFIS) [9] uses a neural network and fuzzy logic to extract fetal ECG. A Kalman filtering-based prediction-estimation technique that uses the current input values and the derived state to estimate the signal’s waveform to estimate fetal ECG [10] . Da Poian et al. [11] used notch filtering and baseline wander removal for pre-processing of ECG recordings. With the help of a compressive sensing sparse binary matrix, a compression ratio of upto 60 % allows extraction of the kind